Images from the MIT-Adobe 5K dataset

First column shows the input images.

Second column shows the retouched images by Expert A.

Third column shows the predicted results for HDRNet.

Fourth and fifth column shows the results for 3 dimensions using PCA and VAE respectively.

Finally we show the results using our method in the last column.

You can toggle between the predicted and the target image by clicking on the images. Moreover, you can rearrange the images in increasing or decreasing order of the metrics.

ID Input Expert A HDRNet PSNR 21.241 FLIP 0.398 Delta E 2000 11.295 PCA PSNR 16.885 FLIP 0.433 Delta E 2000 12.761 VAE PSNR 26.410 FLIP 0.205 Delta E 2000 4.841 Ours PSNR 36.880 FLIP 0.107 Delta E 2000 2.065
000001 19.700 0.380 8.880 17.793 0.352 8.263 22.279 0.301 6.007 42.141 0.064 0.919
000002 22.078 0.397 10.408 21.041 0.365 9.867 32.111 0.120 3.134 34.463 0.098 3.733
000003 23.791 0.209 6.203 8.480 0.580 26.867 32.626 0.109 3.608 41.024 0.076 1.336
000004 22.326 0.300 7.845 19.732 0.272 7.455 27.859 0.148 3.981 40.847 0.056 1.249
000005 27.656 0.242 6.170 13.327 0.501 16.686 20.924 0.304 6.949 39.621 0.069 0.866
000006 26.583 0.295 6.429 20.682 0.351 8.193 27.167 0.193 4.264 41.855 0.072 1.167
000007 22.839 0.312 8.438 13.732 0.572 17.213 27.295 0.176 3.564 30.058 0.159 2.758
000008 21.665 0.374 11.584 19.088 0.364 12.219 23.781 0.230 5.208 30.974 0.141 3.114
000009 29.419 0.267 6.114 23.392 0.250 5.750 31.079 0.163 3.082 55.158 0.026 0.434
000010 25.950 0.240 6.158 22.972 0.256 7.834 30.551 0.144 3.492 38.998 0.085 1.958